Photoaffinity Cross-linking Identifies Differences in the Interactions of an Agonist and an Antagonist with the Parathyroid Hormone/Parathyroid Hormone-related Protein Receptor*

Analogs of parathyroid hormone (PTH)-related protein (PTHrP), singularly substituted with a photoreactivel-p-benzoylphenylalanine (Bpa) at each of the first 6 N-terminal positions, were pharmacologically evaluated in human embryonic kidney cells stably expressing the recombinant human PTH/PTHrP receptor. Two of these analogs, in which the photoreactive residue is either in position 1 or 2 (Bpa1- and Bpa2-PTHrP, respectively) displayed high affinity binding. Bpa1-PTHrP also displayed high efficacy for the stimulation of increased cAMP levels. Surprisingly, Bpa2-PTHrP was found to be a potent antagonist, despite the presence of the principal activation domain (sequence 1–6). Analysis of the digestion profiles of the ligand-receptor photoconjugates revealed that both the agonist and the antagonist cross-link to the S-CH3 group of Met425 in transmembrane domain 6 of the human PTH/PTHrP receptor. However, the antagonist Bpa2-PTHrP also cross-links to a proximal site within the receptor domain Pro415–Met425. Unlike the antagonist Bpa2-PTHrP, the potent agonist Bpa2-PTH, also bearing the Bpa residue in position 2, cross-links only to the S-CH3 group of Met425 (similar to Bpa1-PTHrP and Bpa1-PTH). Taken together, these results suggest that the antagonist Bpa2-PTHrP is able to distinguish between two distinct conformations of the receptor. The comparison between PTHrP analogs substituted by Bpa at two consecutive positions and across PTH and PTHrP reveals insights into the PTH/PTHrP ligand-receptor bimolecular interaction at the level of a single amino acid.

Extensive efforts have been focused on investigating the structural basis for hormone recognition by PTH1R. These lines of research have identified regions in both the ligands and the receptor required for binding and signaling. On one hand, the mutagenesis approach, generating chimeric, mutated, or truncated receptors, has established that multiple receptor domains are involved in the complex interaction with the ligands, including the N-terminal extracellular tail, extracellular loops, and transmembrane domains (TMDs) (9 -11). On the other hand, structure-function studies of the hormones and their analogs have revealed structural determinants required for interaction with the receptor. These studies have identified the C-terminal region of PTH-  and PTHrP-  as the principal binding domain (12)(13)(14), which is also required for activation of protein kinase C (15,16). The N-terminal 1-6 sequence of either ligand was found to function as the principal activation domain (Ref. 17 and references therein).
The complementary biochemical approach of photoaffinity cross-linking has been utilized to examine directly ligand-receptor bimolecular interactions through the generation of covalently linked radiolabeled ligand-receptor photoconjugates. These studies identified three contact sites in PTH1R: position 1 in PTH interacts with Met 425 (18), position 13 in PTH interacts with the sequence [Glu 182 -Met 189 ] (19,20), and position 23 in PTHrP interacts with the sequence 23-40 (21). Thus, the N-terminal part of PTH interacts with TMD 6 of the receptor near the cell surface and extracellular loop 3, the mid-region (position 13) with the receptor at the N-terminal extracellular tail-juxtamembrane region, and position 23 of PTHrP with the extreme N terminus of the receptor. Further data generated by this approach holds the promise of developing an experimentally based model of the hormone-receptor interface (18).
Here, we report the evaluation of a series of photoreactive analogs of PTHrP in order to probe the nature of receptor interaction with the principal activation domain of the hormone (residues 1-6). The receptor "contact domains" for two * This work was supported in part by Grant RO1-DK47940 (to M. R.) from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ This work is presented in partial fulfillment of the requirements of a Ph. D p-benzoylphenylalanine (Bpa)-containing radiolabeled PTHrP analogs, one an agonist and the other a potent antagonist, were identified and compared with the cross-linking sites of the corresponding PTH analogs. The data provide a unique opportunity to obtain insights into potential differences between the modes of interaction of agonists versus antagonists with the PTH1R.
Radioiodinations  34 ]bPTH-(1-34)NH 2 (Bpa 2 -PTH) were radioiodinated and purified by reverse-phase HPLC as described previously (23), but with the following modification: all iodination reactions were continued for 12 min. The iodination reactions of the PTHrP-based analogs were stopped by the addition of acetic acid (final concentration, 20% (v/v)), because the iodinated peptides were found to have low solubility in aqueous solution.
Adenylyl Cyclase Activity-HEK-293/C-21 cells were subcultured in 24-well plates and grown to confluency. Activation of AC by PTHrP analogs was carried out as described (22). Antagonist activity was tested by measuring the inhibition of 50 nM PTHrP-stimulated AC activation following a preincubation period of 15 min with the antagonist at 37°C.
Enzymatic and Chemical Digestions of the Ligand-Receptor Conjugates-Samples of the isolated SDS-PAGE bands representing either the radiolabeled hormone-receptor conjugate or conjugated fragments were prepared in small volumes (typically 10 -20 l) of 25 mM Tris-HCl (pH 8.5) Triton X-100 (0.1% (v/v)), SDS (0.01% (w/v)). Endo-F digestions were carried out at 37°C for 24 h, according to the manufacturer's procedure. Lys-C digestions were performed by treatments with 0.15 units of enzyme (in 10 l water) in 25 mM Tris-HCl (pH 8.5), Triton X-100 (0.1% (v/v)), SDS (0.01% (w/v)) at 37°C for 24 h. BNPS-skatole cleavage was carried out with 100 l of 1 mg/ml (final concentration) in 70% acetic acid at room temperature for 24 h in the dark under N 2 . Samples were dried on a Speed-Vac and dissolved in reducing sample buffer (25) prior to PAGE analysis.
CNBr digestions were performed either in solution or on a solid support (26). In solution, digestions were performed with a small crystal of CNBr in 70% formic acid at 37°C for 24 h in the dark under N 2 (19). On a solid support, samples of the isolated SDS-PAGE bands representing the receptor conjugates were adsorbed onto C 18 -derivatized silica gel (approximately 5000 cpm/10 mg of silica) that was prewashed with 0.1% trifluoroacetic acid in acetonitrile (v/v) (solvent B) followed by 0.1% trifluoroacetic acid in water (v/v) (solvent A). After loading the radiolabeled ligand-receptor conjugate, the silica gel was washed consecutively with 10 gel volumes of solvent A, solvent B, and solvent A. The silica gel was then equilibrated by washing twice with 10 gel volumes of 0.1 M HCl followed by the addition of 2 gel volumes of CNBr to a final concentration of 80 mg/ml in 0.1 M HCl. After overnight incubation at 37°C in the dark under N 2 , the silica gel was washed twice with 10 gel volumes of solvent A to remove the CNBr. CNBrgenerated fragments were eluted from the silica gel with 40 -50% solvent B in A, monitoring the elution profile by the release of the radioactivity from the column.
Electrophoresis and Autoradiography-Electrophoretic analyses were performed using 7.5% (w/v) SDS-PAGE for the intact and deglycosylated hormone-receptor conjugates and 16.5% (w/v) Tricine/SDS-PAGE for the cleavage products (19). Appropriate molecular weight markers (Amersham Pharmacia Biotech) were included in each gel. Gels were dried and exposed to x-ray films (X-omat, Eastman Kodak Co.) with intensifying screens (XAR-5, Kodak). Following autoradiography, the radioactive bands were excised from the dried gels, electroeluted (Bio-Rad, Electroeluter model 422) in SDS-PAGE running buffer, and concentrated on a Speed-Vac.
Receptor Mutagenesis-The generation and biological characterization (binding and AC assays) of two single mutations in the PTH1R cDNA sequence, M414L and M425L, was described previously (18).
Transient Transfection-COS-7 cells were plated at 65,000 cells/well in 24-well dishes, 24 h prior to transient transfections. Six hundred ng of empty vector, mutant, or wild-type receptor cDNA constructs were transfected using 1.8 l of FuGENE TM 6 (Roche Molecular Biochemicals) transfection reagent per well. Photoaffinity cross-linking, radioreceptor binding, and AC activity assays were carried out 48 h after transfection, as described above.

Characterization of Bpa-containing PTHrP-(1-36) Analogs-A series of photoreactive analogs of PTHrP-(1-36)
, singularly substituted with a Bpa at each of the first six Nterminal positions was prepared (Fig. 1A). Lys 11 and Lys 13 in these analogs were replaced by Arg residues to render the ligands resistant to enzymatic cleavage by Lys-C, as required by the mapping scheme. In addition, analogs Bpa 1-4 -and Bpa 6 -PTHrP carry the modification His 5 3 Ile, which markedly enhances PTHrP interaction with the type 2 PTH receptor (PTH2R) subtype (27,28). Preliminary experiments estab- (1-34)NH 2 in receptor binding and AC activation assays (data not shown).
Substitution at position 3 led to an analog with a markedly reduced binding affinity (IC 50 ϭ 310 nM), whereas substitution at positions 4 and 5 essentially abrogated binding.
Identification of the Cross-linking Contact Sites of the Agonist 125 I-Bpa 1 -and the Antagonist 125 I-Bpa 2 -PTHrP on PTH1R-The ϳ90-kDa bands corresponding to either 125 I-Bpa 1 -or 125 I-Bpa 2 -PTHrP-PTH1R conjugates were isolated from 7.5% SDS-PAGE and subjected to a series of chemical and enzymatic cleavages. One digestion pathway consisted of enzymatic digestion with Lys-C at the carboxyl side of lysyl residues, followed by chemical cleavage with BNPS-skatole at the carboxyl side of tryptophanyl residues. Exhaustive Lys-C treatment of the ϳ90-kDa ligand-receptor conjugates of either 125 I-Bpa 1 -or 125 I-Bpa 2 -PTHrP yielded single radiolabeled bands with similar apparent molecular masses of ϳ12 kDa, as analyzed by 16.5% Tricine/SDS-PAGE (Fig. 3, A, lane 1, and C, lane 1, for 125 I-Bpa 1 -and 125 I-Bpa 2 -PTHrP, respectively). Similar treatment of the deglycosylated conjugates (ϳ60 kDa) yielded fragments with the same apparent molecular masses (not shown), indicating the absence of glycosylation sites within the Lys-C-generated fragment. BNPS-skatole treatment of the excised and eluted ϳ12-kDa bands produced a single band for each radioligand with similar apparent masses of ϳ7-8 kDa (Fig. 3, A, lane 2, and C, lane 2, for 125 I-Bpa 1 -and 125 I-Bpa 2 -PTHrP, respectively).
The second digestion pathway, the reciprocal of the first one, yielded upon BNPS-skatole treatment of each of the intact hormone-receptor conjugate a band migrating at ϳ13-14 kDa (Fig. 3, A, lane 3, and C, lane 3, for 125 I-Bpa 1 -and 125 I-Bpa 2 -PTHrP, respectively). Lys-C treatment of the isolated ϳ13-14-kDa bands yielded ϳ7-8-kDa fragments for either conjugate, similar to the conjugated fragments obtained from the first digestion pathway (Fig. 3, A,  The isolated radioactive bands corresponding to the intact hormone-receptor conjugates were treated also with cyanogen bromide (CNBr). For 125 I-Bpa 1 -PTHrP-PTH1R conjugate, this treatment generated a single band with an apparent mass of ϳ4.5 kDa (Fig. 3B, lane 1), similar to the gel mobility of the free ligand 125 I-Bpa 1 -PTHrP (Fig. 3B, lane 2). CNBr treatment of the 125 I-Bpa 2 -PTHrP-PTH1R conjugate yielded two bands with apparent masses of ϳ6 and ϳ4.5 kDa (Fig. 3D). The later band had a gel mobility similar to the band obtained by CNBr treatment of the 125 I-Bpa 1 -PTHrP-PTH1R conjugate. The size of the ϳ6-kDa band was not further reduced by retreatment with CNBr to achieve exhaustive digestion, suggesting that this band represents the minimal sized CNBr-generated fragment. In addition, subsequent treatment with either Lys-C or BNPSskatole did not reduce the size of this fragment, suggesting the absence of cleavage sites for these reagents (not shown).
These mutated receptors maintain biological properties similar to the wild-type receptor when transiently expressed in COS-7 cells (18): the AC activation profiles of wild-type receptor, mutant M414L, and mutant M425L in response to PTHrP-(1-36) were comparable (Fig. 4A). The antagonistic activity of Bpa 2 -PTHrP was observed in transiently expressed native and mutated receptors (Fig. 4B); the dose-inhibition of PTHrP-(1-36)stimulated AC curves were similar to those obtained in the HEK-293/C-21 cells stably expressing PTH1R (Fig. 1D).
Cross-linking of 125 I-Bpa 2 -PTH to PTH1R generated a radiolabeled band migrating at ϳ90 kDa, corresponding to the intact hormone-receptor conjugate (not shown). Subsequent CNBrcleavage yielded a single band migrating at ϳ4.5 kDa (Fig. 5A,  lane 2), similar to the mobility of the ligand 125 I-Bpa 2 -PTH (molecular weight, 4460) (Fig. 5A, lane 3). This result therefore suggests that all three agonists, Bpa 1 -PTHrP, Bpa 1 -(18), and Bpa 2 -PTH, interact with PTH1R in a very similar manner. In order to analyze the receptor interactions of 125 I-Bpa 2 -PTH in greater details, we cross-linked it to the mutants M414L and M425L.
Photoaffinity cross-linking of radioiodinated Bpa 1 -and Bpa 2 -PTHrP to PTH1R followed by consecutive and reciprocal treatments with Lys-C and BNPS-skatole generated similar digestion patterns for the two ligand-receptor conjugates. These treatments yielded bands with similar electrophoretic mobilities at ϳ7-8 kDa that are not affected by Endo-F treatment (Fig. 3, A and C). The molecular weights of 125 I-Bpa 1 -and 125 I-Bpa 2 -PTHrP are 4645 and 4618, respectively, and therefore, the receptor fragment contributing to the final ϳ7-8 kDa of the photoconjugated fragments is approximately 3.5 kDa. Examination of the theoretical Lys-C and BNPS-skatole digestion maps of PTH1R reveals only one possible nonglycosylated fragment consistent with the observed fragmentation. This fragment corresponds to Ser 409 -Trp 437 , which includes most of TMD 6 and part of extracellular loop 3 (Fig. 6A).
CNBr digestion of the 125 I-Bpa 1 -PTHrP-PTH1R conjugate (Fig. 3B) suggests that cross-linking occurred at the ⑀-methyl of a Met residue. In such a case, CNBr treatment would yield a conjugated fragment that is simply a "CH 2 SCN"-modified ligand represented by the ϳ4.5-kDa band and therefore indistinguishable from ligand alone by SDS-PAGE (18,26). CNBr digestion of the 125 I-Bpa 2 -PTHrP-PTH1R (antagonist) conjugate suggests that cross-linking occurred at two distinct sites (Fig. 3D). One site is the ⑀-methyl of a Met residue, similar to Bpa 1 -PTHrP cross-linking, and the other is a different site, as evidenced by the additional CNBr-generated fragment of ϳ6 kDa (Fig. 3D). The additional site could be either a different amino acid or the ␥-CH 2 within the same Met, which would result in the generation of a fragment of ϳ6 kDa. The receptor fragment contributes 1-2 kDa to the ϳ6-kDa fragment conjugate, which corresponds to either Pro 415 -Met 425 or Ala 426 -Met 441 . The finding that the size of the ϳ6-kDa fragment is not further reduced following treatment with either Lys-C or BNPS-skatole indicates that Pro 415 -Met 425 is the minimal digestion-restricted domain that includes the photoinsertion site for 125 I-Bpa 2 -PTHrP (Fig. 6B).
The contact domain identified by Lys-C and BNPS-skatole digestions contains 2 methionine residues, Met 414 and Met 425 , either of which could be the putative "contact point" for 125 I-Bpa 1 -PTHrP or 125 I-Bpa 2 -PTHrP. Photoaffinity cross-linking to single point-mutated receptors (M414L and M425L) suggests that the benzophenone moiety in both the agonist Bpa 1and the antagonist Bpa 2 -PTHrP contacts Met 425 in TMD 6 of PTH1R. Cross-linking of 125 I-Bpa 2 -PTHrP to M425L mutant receptor yielded a very weak but specifically labeled radioactive band at ϳ90 kDa (Fig. 4D, lane 5), whereas no crosslinking was detected with radioiodinated Bpa 1 -PTHrP. This observation suggests that cross-linking of the antagonist to the M425L mutant does occur, but with a markedly reduced efficiency. The low level of incorporation into the M425L receptor mutant precludes detailed analysis of the actual contact domain. This finding, however, emphasizes a distinct difference between the cross-linking of 125 I-Bpa 1 -and 125 I-Bpa 2 -PTHrP to PTH1R and corroborates the CNBr digestion data.
The results may reflect either differences between the binding modes of the agonist and the antagonist or differences in the interaction between the two consecutive positions in the PTHrP-(1-36) sequence and PTH1R. In an attempt to distinguish between these two possibilities, we utilized the agonist analog Bpa 2 -PTH, which carries the same photoreactive moiety at the same position as the antagonist Bpa 2 -PTHrP. Analysis of 125 I-Bpa 2 -PTH photoconjugates with wild-type, M414L, and M425L mutated PTH1R indicates that this ligand cross-links only to the ⑀-methyl of Met 425 (Fig. 5), similar to Bpa 1 -PTHrP (Fig. 3, A and B, and Fig. 4C) and to Bpa 1 -PTH cross-linking (18). These results, therefore, provide strong support for the hypothesis that the differences observed between the crosslinking of 125 I-Bpa 1 -and 125 I-Bpa 2 -PTHrP may reflect different interaction modes of an agonist versus an antagonist with the receptor.
One of the models of GPCR activation suggests that in the absence of a ligand, the receptor is in equilibrium among several conformational states, spanning the transition between resting and activated conformations (32). According to this model, antagonists bind the receptor without affecting this equilibrium, whereas agonist interaction shifts the equilibrium to the activated conformation. One possible interpretation of our results is that in the "agonist-bound" conformation (activated), the photoinsertion site for either Bpa 1 -or Bpa 2 -PTHrP is the ⑀-methyl of Met 425 . In the presence of antagonist, where a more heterogeneous population of antagonist-receptor complexes prevails, there are receptor conformations in which the photophore at position 2 is positioned to allow photoinsertion to an alternate site or several other distinct sites in proximity to Met 425 . Hence, the additional cross-linking site of the antagonist Bpa 2 -PTHrP (represented by the ϳ6-kDa band) is probably obtained through interaction with receptors in a conformation different from the one obtained in the presence of the agonist.
Several publications have discussed the binding modes of agonists versus antagonists in the GPCR superfamily for the following receptors in particular: neurokinin 1 (33)(34)(35), cholecystokinin B/gastrin (36,37), angiotensin-1 (38), corticotropinreleasing factor (39), B 2 bradykinin (40), neuromedin B (41), and gonadotropin-releasing hormone (42). These studies have concluded that the binding sites of peptide agonists are distinct from those of peptide or nonpeptidic antagonists. In contrast, a few other investigators have found that peptide agonists and peptide or nonpeptidic antagonists have overlapping binding sites on their receptors, including the B 2 bradykinin (43), the ␤-adrenergic (44), the cholecystokinin (45), and the neuropeptide Y Y1 receptors (46). These differences may be attributed to elucidation of only a subset out of multiple interaction sites between ligand and receptor, especially in the case of intermediate-size peptide ligands. It is conceivable that some interaction sites are distinct whereas some others overlap, particularly, as presented here, if the agonist and antagonist are structurally very similar.
The finding that both residues 1 and 2 in both PTH-(1-34) and PTHrP-(1-36) cross-link to Met 425 (or a larger segment in TMD 6) is consistent with the idea that this region of the receptor is essential for activation. TMD 6 is contiguous with intracellular loop 3, which is hypothesized to be coupled to G s (47,48). Although the molecular details of activation of GPCRs are not fully understood, it has been postulated that, upon agonist binding, conformational changes are induced in the receptor that affect the intracellular loop domains, such as intracellular loop 3, to increase affinity for G proteins (guanyl nucleotide-binding proteins). Using site-selective fluorescently labeled ␤2 adrenergic receptor, experimental evidence suggested that movements of TMD 3 and 6 underlie the activation of the receptor (49). Our observations are in line with this hypothesis and suggest that TMD 6 is involved in activation of PTH1R, and probably of the entire PTH/secretin subfamily.
The finding that Bpa 2 -PTHrP is an antagonist, whereas Bpa 2 -PTH is an agonist, suggests that PTH-(1-34) and PTHrP-(1-36) have some distinct sets of interactions with their common receptor. Moreover, PTH1R is able to distinguish between PTH-(1-34) and PTHrP-(1-36) analogs substituted at position 2 by Bpa (and presumably other amino acids). The nonequivalence of PTH and PTHrP is demonstrated by the selective interaction of PTH-(1-34) with the PTH2R, which does not recognize PTHrP-(1-36) (50). Replacement of His 5 by Ile in PTHrP-(1-36) overcomes this distinction and converts the analog into a potent agonist (27,28). Interestingly, we found that Bpa 2 -PTHrP the analog is a full agonist for the PTH2 receptor. 2 This observation suggests that Bpa 2 -PTHrP is able to distinguish between two highly homologous receptor subtypes and emphasizes the key role of the interaction between position 2 and the receptor. The critical role of this position is also illustrated by the [Arg 2 ]hPTH-(1-34) analog, which interacts differently with PTH1R from different species (30).
In conclusion, this study examined the cross-linking sites of peptide agonist and antagonist analogs of PTHrP. We found that both Bpa 1 -and Bpa 2 -PTHrP cross-link to the ⑀-methyl of Met 425 in TMD 6 of PTH1R. The antagonist Bpa 2 -PTHrP also cross-links to a distinct proximal region, probably within the domain Pro 415 -Met 425 . These differences in the cross-linking sites are attributed to different interactions of peptide agonists and antagonists with the PTH1R and suggest that the antagonist Bpa 2 -PTHrP interacts with distinct populations of the receptor, one or more of which is distinct from the conformation recognized by the agonist.